A growing body of evidence supports a characterization of reading disability (RD) as a brain-based difficulty in acquiring fluent reading skills, usually associated with phonological processing deficits (Lyon et al., 2003). At a cognitive level, these deficits are thought to impede adequate binding of those orthographic (O), phonological (P), and semantic (S) features that form the basis of lexical representations necessary to support efficient reading (cf. Harm & Seidenberg, 1999; Perfetti & Hart, 2001). At the level of brain systems, neuroimaging studies have consistently shown that, relative to typically developing (TD) readers, RD readers show both structural and functional differences at those left hemisphere (LH) regions that comprise the distributed circuitry for O, P, and S processing. At present though, we do not know which of these findings of structural and functional anomalies are causally related to specific learning problems in the domain of reading and which are not. A crucial next step is to generate and test predictive models of individual differences in O, P and S learning under controlled experimental conditions. To that end, we will employ integrated neurobiological and behavioral measures during novel spoken and written word learning tasks to determine which structural and functional brain differences predict word learning difficulties in general, and whether the strongest brain predictors vary as the relative demands on O, P or S coding are manipulated. Also, despite the general consensus that the primary deficit in RD arises at the level of decoding and phonological processing skills, there are clinical, behavioral, and neuroimaging findings which suggest additional problems in consolidation of newly learned information into permanent memory in RD. To explore this hypothesis, we will employ integrated brain/behavior analyses to examine both online learning and offline consolidation of newly learned O, P and S forms. To examine neural mechanisms underlying consolidation deficits more precisely we will also use targeted linguistic and non-linguistic learning tasks that systematically modulate demands on procedural vs. declarative memory systems. For all of these language and non-linguistic measures of learning and consolidation we test specific brain-based models of TD/RD differences.